Horizon Accord | Fertilizer | Hormuz | Machine Learning
Where the Fertilizer Shock Went
In April, this publication argued that a fertilizer disruption at the Strait of Hormuz would move through the food system quietly, without a single headline moment. Five months of trade data, farm surveys, USDA records, and livestock economics now let that pathway be traced with more precision than was possible in April: real, measurable stress at the farm-input level, and — so far, in the data available — limited evidence that it has traveled further than that. The Strait remains functionally closed as of this writing, and that qualifier matters.
What Still Holds
The April 4 argument rested on a structural observation, not a forecast, and the structural observation has held up under five months of scrutiny. The Strait of Hormuz carries a large share of the world's seaborne nitrogen fertilizer trade. The spring application window is time-sensitive in a way few other agricultural inputs are — miss the relevant application window, and later shipments cannot simply restore the lost opportunity. And no coordinated international fertilizer emergency architecture exists comparable to the International Energy Agency's petroleum reserve system. Carnegie Endowment made the same argument independently in March. The Food and Agriculture Organization made it in its own words weeks after publication. By August, a JPMorgan research note was describing the identical asymmetry — a fertilizer buffer the West does not have, while China stockpiles one. That convergence is the strongest evidence the original thesis has received.
The April piece also sketched, without treating it as a firm mechanistic model, a plausible downstream pathway under real uncertainty: shortage to cost to yield to price to plate. Five months of primary data — pulled from USDA, university agricultural economists, and the fertilizer industry's own earnings calls — now let that pathway be resolved with more precision than the uncertainty of April allowed. This piece traces where the pressure is documented clearly, where it thins out, and where the record simply does not yet reach.
A Nitrogen Shock, Not a Fertilizer Shock
"Fertilizer crisis" is the wrong scale for what happened. University of Illinois farmdoc economists tracked Illinois input prices through the disruption and found anhydrous ammonia rising from a pre-conflict average of $828 a ton to $1,123 by mid-April — a 36 percent jump — with 28 percent UAN solution up 25 percent over the same window. Diammonium phosphate and potash barely moved: DAP rose less than one percent, potash about two percent. The reason is mechanical rather than geopolitical. Most phosphate and potash for the 2026 crop had already been bought and applied the previous fall, before the conflict began, and potash in particular is far less dependent on Gulf shipping to begin with.
The distinction matters because it changes what evidence should be expected downstream. A genuinely broad, multi-nutrient fertilizer shock should leave a wider nutrient and crop signature than the strongly nitrogen-concentrated pattern actually observed. A nitrogen-specific shock concentrates its force on corn and wheat — where nitrogen can run a third or more of per-acre operating cost — and largely spares soybeans, which fix their own nitrogen from the atmosphere and draw ten percent or less of their cost from fertilizer at all. Any comparison between corn and soybean outcomes this season has to start from that asymmetry, not treat it as a surprise finding.
A Shock That Landed on Already-Weak Ground
Purdue's own farm-policy briefing in March 2026 noted that USDA's February baseline — published before the Strait closed — had already begun projecting a shift from corn to soybeans, because the nitrogen-to-corn price ratio was, in the brief's words, "near historically poor levels" independent of anything happening in the Persian Gulf. Potash prices were separately up 21 percent year over year as early as September 2025, months before the war began, driven by tariff concerns on Canadian imports rather than Hormuz. The nitrogen-to-corn price ratio itself — a standard measure of how many bushels of corn it takes to pay for a unit of fertilizer — sat near 0.11 as a multi-year baseline before climbing toward 0.20 during the crisis. That exceeds the roughly 0.17 peak reached during the 2022 Russia-Ukraine shock, by this measure.
What made 2026 more damaging to farm margins than 2022's comparable price ratio, according to multiple agricultural economists, was the absence of an offsetting factor: in 2022, high fertilizer prices arrived alongside record-high corn prices, which cushioned the blow. In 2026, corn prices stayed low through the entire disruption window — a point verified directly against USDA's own farmer-price records below. The Hormuz shock did not create farm-economic stress from a neutral starting point. It accelerated a squeeze that was already underway and removed the one variable that had softened the equivalent shock four years earlier.
What Actually Moved at the Farm Gate
The American Farm Bureau Federation's April survey found 70 percent of U.S. farmers could not afford the full fertilizer their operations needed, with more than 80 percent of rice, cotton, and peanut growers reporting the same shortfall against roughly half of Midwest row-crop producers. The gap tracked pre-booking behavior almost exactly: soybean and barley growers, who lock in fertilizer earliest, reported the least exposure; cotton and peanut growers, who pre-book at the lowest rates of any surveyed crop, reported the most. Individual accounts matched the pattern at the ground level — a Screven County, Georgia, row-crop farmer told Georgia Farm Bureau that a planned corn-acreage increase reversed into a roughly 40 percent cut once fertilizer prices moved from 55 to 96 cents a pound within weeks, with the freed acreage shifted to soybeans as the only crop that still penciled out.
Where the money actually went is the least examined part of the story until now. CF Industries' first-quarter 2026 EPS rose 115 percent year over year; Yara International's second-quarter EBITDA rose 39 percent to its highest quarterly level in a decade. A StoneX analyst tracking the market noted that U.S. nitrogen producers' own input costs "remain relatively stable" even as their sale prices spiked, because much of their natural gas feedstock is domestic rather than Gulf-sourced. The price American and Brazilian farmers paid did not simply reflect a proportional rise in producers' own costs: producer margins expanded while at least some producers' own feedstock costs stayed comparatively stable. What share of the farmer-level price increase this accounts for cannot be quantified from these figures alone, but the direction is documented in the companies' own disclosures.
Yara also disclosed something harder to interpret cleanly: crop-nutrition delivery volumes fell 17 percent year over year in the second quarter, with management attributing $240 million of the quarter's impact to lost volume. That is the closest thing to a hard consumption signal found anywhere in five months of research — and it cuts in an ambiguous direction. It could reflect farmers genuinely reducing the nitrogen they apply, which is consistent with the causal chain the original piece described. It could equally reflect delayed purchasing, drawn-down dealer inventory, or ordinary seasonal timing quirks that a single quarter's volume figure cannot distinguish. It should be read as evidence that something changed in farmer purchasing behavior, not as confirmation of reduced field-level application.
What the price record shows
Anhydrous ammonia: $828/ton (pre-conflict average) → $1,123/ton by mid-April 2026, +36%. Source: farmdoc daily, University of Illinois.
DAP: $862/ton → $870/ton over the same window, +1%. Potash: $493/ton → $505/ton, +2%. Same source.
Nitrogen-to-corn price ratio: ~0.11 multi-year baseline → ~0.20 by April 2026. Source: farmdoc daily and Michigan State University Extension.
CF Industries Q1 2026 EPS: up 115% year over year. Yara Q2 2026 EBITDA: up 39%, highest quarterly figure in a decade. Source: company filings, reported via Investing.com.
The Crop Story Doesn't Resolve Cleanly
USDA's August Crop Production report cut the 2026 U.S. corn yield estimate to 180.7 bushels an acre, down from a record 186.5 in 2025, on harvested acreage of 88.6 million acres, down from 91.3 million the year before. Agricultural trade press attributed the yield cut chiefly to western crop stress, hail, and storm losses documented in USDA's own crop-condition ratings — not to fertilizer application rates. That attribution is repeated more often than it is demonstrated: no source located in this research isolates a fertilizer-specific yield effect from a weather effect at the county level, and USDA's own thirty-year accuracy record shows August estimates carry a typical margin of error exceeding five bushels an acre, meaning the current number should not be treated as final.
The regions where this is hardest to untangle are the same regions twice over. The South and Southeast reported both the worst fertilizer affordability of any region surveyed by the Farm Bureau and sustained heat and dryness through late summer, according to USDA's own weekly crop bulletins. A state-level statistical model built on thirty years of crop-condition data found Tennessee's actual 2026 corn yield coming in nearly 25 bushels an acre below what condition ratings alone predicted, despite the model's unusually strong historical fit for that state. A separate North Dakota State University drought-projection model, checked against the same anomaly, did not cleanly explain it either — Tennessee does not appear among the states that model flags for significant 2026 drought exposure. That rules out one easy explanation without supplying a replacement. Nobody has isolated why. It is the clearest quantified anomaly in the record, and it should stay exactly that: an anomaly, not evidence.
Acreage is the cleaner half of the story, and even it is not entirely clean. On a like-for-like basis, U.S. corn harvested acreage fell from 91.3 million in 2025 to 88.6 million in 2026 — a real decline, though a smaller one than mixing planted and harvested figures across the two years would suggest, and it moves in the direction the affordability data predicted. But Purdue's own pre-conflict forecasting had already flagged a corn-to-soybean shift before Hormuz closed, meaning part of this year's acreage movement continues a trend that predates the crisis rather than originating from it.
Why Less Nitrogen Doesn't Automatically Mean Less Corn
A North Dakota State University white paper published in August, examining the crop-insurance system's interaction with the fertilizer shock, cited established agronomic literature finding that U.S. corn producers typically apply nitrogen above the profit-maximizing rate as insurance against uncertain soil nitrogen supply — and that reductions of 12 to 16 percent are agronomically feasible without measurable expected yield loss under normal conditions. This is not a claim manufactured to fit a thesis. Extension agronomists have described the standard "yield goal" approach to nitrogen as producing systematic over-fertilization for the same reason for decades, and peer-reviewed work on corn nitrogen response finds that the economically optimal rate varies year to year primarily with rainfall during a specific window of the growing season — meaning weather governs not the insurance logic itself, but whether reducing that buffer actually costs yield in a given season.
The practical implication cuts against a simple version of the original causal chain. Less nitrogen does not mean less yield in a season with normal or favorable rainfall at the right growth stages — it means a thinner buffer if conditions turn adverse at the wrong moment. Whether 2026 was that kind of season, in the specific places where farmers actually reduced application, is a question the current record cannot answer, because no measured nitrogen-consumption data for 2026 yet exists in any public dataset. USDA and industry consumption reports lag the crop year by months to years, and every farmer-behavior claim in five months of reporting on this crisis — including in this publication's own prior coverage — has rested on survey intentions and price-ratio modeling rather than measured application.
Where the Livestock Signal Is Missing
The April piece did not make a livestock effect part of its original argument. This publication added the livestock-feed pathway to this investigation because it is the obvious next link a nitrogen-cost shock would have to cross to reach meat, dairy, and eggs — and because testing it turned up one of the clearer results in five months of research. USDA's own monthly farmer-price data shows corn selling for less in every month of 2026 through July than in the same month of 2025, including through March-to-July, the window that follows the Strait's closure. Soybean meal, the other dominant livestock feed input, was forecast through 2026 in the $300-to-$317-a-ton range, below its 2007-to-2025 long-run average of $360, with USDA attributing the movement to record Brazilian production and ample global stocks rather than any input-cost pressure.
Every livestock sector examined shows the same pattern. Broiler chick placements ran above trend through nearly all of 2026, with feed costs explicitly described by industry analysts as "subdued." Egg-layer flocks expanded 4.8 to 5.3 percent year over year, and the dominant 2026 price story in that sector was a July avian-influenza outbreak in Utah — a disease event, not a feed-cost event. The U.S. dairy herd reached its largest size since 1994, with Purdue's own dairy-market analysis attributing 2026 margin pressure to falling milk prices rather than rising feed costs, "even with sustained cheaper feed inputs." Cattle inventory data reflects the tail of a twelve-year cyclical contraction that predates 2026 and appears to be stabilizing, with heifer retention rising rather than falling.
None of this means the farm-level pressure documented earlier in this piece was not real. It means the pressure has not yet crossed the specific threshold — a measurable rise in the price of corn or soybean meal — that would be required to reach a hog, a broiler, a dairy cow, or a steer. The downstream pathway being tested here has a missing link, and the missing link is not obscure: it is the single most-tracked commodity price in American agriculture, and it has not moved in the direction that pathway would require.
What Prior Shortages Say About Where the Cost Goes
Three prior episodes offer a rough guide, and each shows absorption rather than clean transmission. After the 1973 oil embargo triggered a real, documented decline in fertilizer application rates, retail food prices were already rising sharply — 20 percent annualized in the first quarter of 1974 alone. The larger second-half acceleration, 13.4 percent annualized from June to December, was attributed at the time to a specific additional cause: an anticipated bumper harvest that failed to materialize because of separate, unrelated poor weather, not to the fertilizer shortage itself. Substantial food inflation preceded that second wave; the second wave had its own distinct trigger. A Congressional Research Service review of the 2007-08 fertilizer and grain-price spike found that "rising costs have largely been absorbed by livestock and poultry producers, often with significant financial loss" before reaching consumers, with dairy specifically insulated because much of its feed was home-grown or contracted in advance. And a 2024 peer-reviewed retrospective on the 2021-22 Russia-Ukraine fertilizer shock concluded that the era's worst fears about global consumption collapse "have not come to pass," because trade rerouted around the disruption more effectively than anticipated.
What is genuinely different about 2026, compared with those three episodes, is the nature of the disruption itself. Gulf fertilizer is physically bottlenecked at a single chokepoint rather than restricted by a reroutable trade policy, which historical precedent suggests should make substitution harder than in 2022. It is also arriving on top of a U.S. nitrogen-production base that is now substantially insulated from Gulf natural gas costs — a structural feature that did not exist in 1973 or 2007, and one that may explain why U.S. producer margins expanded even as global fertilizer prices spiked.
What the Record Shows
The structural vulnerability the original piece identified in April was real and remains real: no coordinated fertilizer emergency architecture exists comparable to the one built for oil, and that gap sat exposed for five weeks before this publication named it. Five months on, the clearest evidence available shows that gap producing real, measurable stress at the farm-input level — affordability, substitution, acreage decisions, and a documented shift of value toward fertilizer producers whose own costs did not rise in step. Downstream transmission beyond the farm-input level remains limited in the data available so far: no measurable effect has yet appeared in crop prices at the farm gate, feed costs, or any livestock sector examined.
That is not the same as saying the story is over. The fall 2026 fertilizer-purchasing season, when North Dakota State University's modeling projects urea could hold above $700 a short ton through November, is the point at which the next crop year's exposure gets locked in — and in both 1974 and 2022, the real test of a shock showed up in the following year's decisions, not the year of the shock itself. Brazil's 2026-27 soybean season, whose purchasing window landed squarely inside the disruption in a way the U.S. season did not, will not report a harvest outcome until sometime in 2027. The question this record leaves open is no longer whether the disruption mattered. It is whether the buffers that absorbed it this year — cheap feed grain, expanding herds, producer margin, substitutable nitrogen sources, a favorable weather year in most of the Corn Belt — hold again if a second shock, weather or otherwise, arrives on top of this one before the first has finished working its way through.
Conclusion
Five months later, the record does not show that the warning was misplaced. It shows that the system absorbed more of the first shock than could be seen in April.
That matters. Farmers absorbed some of the cost. Producers captured some of it. Buyers substituted, delayed, or reduced demand. Cheap feed grain prevented the livestock chain from activating. In much of the Corn Belt, weather did not deliver the kind of second shock that has historically turned fertilizer stress into a broader food-price event.
Those are buffers. They are not guarantees.
That is why preparedness was — and remains — the appropriate response. The point of urging households, congregations, food banks, and community organizations to build additional margin was never that famine was certain or that grocery shelves were about to empty. It was that a critical agricultural input had been disrupted at a major global chokepoint, there was no coordinated fertilizer emergency reserve to replace it, and the eventual consequences could not yet be known.
We now know more. The immediate pressure has concentrated primarily at the farm-input level rather than moving cleanly through the food chain. That is good news. It does not erase the vulnerability that produced the warning.
The Strait remains functionally closed. The next crop cycle is beginning to price and purchase its inputs while that disruption is still active. The prudent conclusion is therefore not to declare the danger realized, nor to declare it passed. It is to keep the additional capacity we built, keep watching the places where the pressure is accumulating, and be grateful that preparedness has so far had more time than necessity.
